appsettings.json

{
  "Logging": {
    "LogLevel": {
      "Default": "Warning",
      // Examples: how to handle logs differently by class
      //      "Microsoft.KernelMemory.Handlers.TextExtractionHandler": "Information",
      //      "Microsoft.KernelMemory.Handlers.TextPartitioningHandler": "Information",
      //      "Microsoft.KernelMemory.Handlers.GenerateEmbeddingsHandler": "Information",
      //      "Microsoft.KernelMemory.Handlers.SaveEmbeddingsHandler": "Information",
      //      "Microsoft.KernelMemory.DocumentStorage.AzureBlobs": "Information",
      //      "Microsoft.KernelMemory.Pipeline.Queue.AzureQueues": "Information",
      "Microsoft.AspNetCore": "Warning"
    }
  },
  "KernelMemory": {
    "Services": {
      "AzureOpenAIText": {
        // "ApiKey" or "AzureIdentity"
        // AzureIdentity: use automatic Entra (AAD) authentication mechanism.
        //   You can test locally using the AZURE_TENANT_ID, AZURE_CLIENT_ID, AZURE_CLIENT_SECRET env vars.
        "Auth": "AzureIdentity",
        // Optional when Auth == AzureIdentity. Leave it null to use the default.
        // When the service is on sovereign clouds, this setting might be necessary to configure Entra auth tokens.
        // - https://github.com/Azure/azure-sdk-for-net/blob/main/sdk/openai/Azure.AI.OpenAI/README.md
        // - https://github.com/Azure/azure-sdk-for-net/blob/main/sdk/openai/Azure.AI.OpenAI/src/Custom/AzureOpenAIAudience.cs
        "AzureIdentityAudience": null,
        "Endpoint": "https://<...>.openai.azure.com/",
        "APIKey": "",
        "Deployment": "",
        // The max number of tokens supported by model deployed
        // See https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models
        "MaxTokenTotal": 16384,
        // "ChatCompletion" or "TextCompletion"
        "APIType": "ChatCompletion",
        // How many times to retry in case of throttling.
        "MaxRetries": 10
      },
      "OpenAI": {
        // Name of the model used to generate text (text completion or chat completion)
        "TextModel": "gpt-4o-mini",
        // The max number of tokens supported by the text model.
        "TextModelMaxTokenTotal": 16384,
        // What type of text generation, by default autodetect using the model name.
        // Possible values: "Auto", "TextCompletion", "Chat"
        "TextGenerationType": "Auto",
        // Name of the model used to generate text embeddings
        "EmbeddingModel": "text-embedding-ada-002",
        // The max number of tokens supported by the embedding model
        // See https://platform.openai.com/docs/guides/embeddings/what-are-embeddings
        "EmbeddingModelMaxTokenTotal": 8191,
        // OpenAI API Key
        "APIKey": "",
        // OpenAI Organization ID (usually empty, unless you have multiple accounts on different orgs)
        "OrgId": "",
        // Endpoint to use. By default the system uses 'https://api.openai.com/v1'.
        // Change this to use proxies or services compatible with OpenAI HTTP protocol like LM Studio.
        "Endpoint": "",
        // How many times to retry in case of throttling
        "MaxRetries": 10,
        // The number of dimensions output embeddings should have.
        // Only supported in "text-embedding-3" and later models developed with
        // MRL, see https://arxiv.org/abs/2205.13147
        "EmbeddingDimensions": null,
        // How many embeddings to calculate in parallel.
        // See https://platform.openai.com/docs/api-reference/embeddings/create
        "MaxEmbeddingBatchSize": 100
      }
    }
  }
}